Hi Paul, The change to use the MixinVector is to avoid using multiple inheritance with std::vector, as std::vector doesn't have a virtual destructor the destruction of the vector is undefined w.r.t the C++ Standard. What the MixinVector does is to change the setup so that classes that subclass from MixinVector "have a" std::vector, rather than inherit from std::vector.
The MixinVector is still uses std::vector<> internally, and delegates all functionality to std::vector<> via inline methods. So the OSG itself isn't moving away from STL in any way, just working with it more correctly/robustly. Robert. On Thu, Jun 26, 2008 at 3:55 PM, Paul Martz <[EMAIL PROTECTED]> wrote: > Can someone please explain the MixinVector change and why it was necessary? > Also, please explain the name "MixinVector" -- what, exactly, is a "Mixin"? > If this is a design pattern, I must admit I'm not familiar with it. > > I checked the archives and I see no discussion of this change, not even the > submission. I see only Robert's change log post when requesting 2.5.3 > testing, which shows this change was added on June 19. > > I can see that the change apparently fixes some memory leaks. Are there test > cases that reproduce the problem? If so, can someone post them? > > If it seems that I'm overly concerned, it is because I feel that one of > OSG's strengths is its reliance on STL. So, changes that move OSG away > from STL worry me a bit. This change replaces a well-tested std::vector with > a completely new MixinVector and therefore invalidates all testing on OSG's > Array classes done over the past 10 years. > > Paul Martz > Skew Matrix Software LLC > http://www.skew-matrix.com > +1 303 859 9466 > > _______________________________________________ > osg-users mailing list > [email protected] > http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org > > _______________________________________________ osg-users mailing list [email protected] http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
